Correlated variance in simultaneous inductively coupled plasma atomic-emission spectrometry: its causes and correction by a parameter-related internal standard method

Abstract

For practical purposes the precision of inductively coupled plasma atomic-emission spectrometry (ICP-AES) is limited by medium-term changes of sensitivity, i.e., those occurring on a time scale shorter than the period between calibration adjustments. For any ICP-AES it is this longer term lack of reproducibility that limits its capacity to match other multi-element methods (such as X-ray fluorecence) for high precision analysis.

This paper describes the application of principal components analysis to show that more than 90% of the total variance in routine analysis by ICP-AES is not random but multi-element correlated and therefore potentially correctable. Ten possible causes of this variation are quantified in terms of their particular multi-element respose “signatures”. Comparison of the ten signatures suggests that only two signatures predominate, despite being produced by seven apparently independent instrumental parameters.

A new method of improving the medium-term precision of ICP-AES is described in which the variation from the two predominating signatures on all analytes is removed by use of the parameter-related internal standard method (PRISM). Two internal standard elements are chosen each of which responds significantly to only one of the two predominating signatures. Two instrumental parameters, forward power and nebuliser uptake rate, are varied separately. In each instance the response of each analyte is related to corresponding variation of the appropriate internal standard. For subsequent routine analysis a straightforward mathematical correction is then applied to each analyte using the measured function relating that analyte with both internal standards.